Numerical weather prediction models
The Unified Model is run operationally in a number of configurations for weather forecasting at the Met Office.
Global and regional configurations
Global configurations of the Unified Model provide the medium-range weather forecasts and also support the nested higher resolution regional models with boundary data. More detailed short- to medium-range forecasts are provided by these high-resolution models which are able to represent certain atmospheric processes more accurately, as well as having a more detailed representation of surface features such as coastlines and orography.
The deterministic UK forecast uses a variable resolution model which has a high resolution inner domain (1.5 km grid boxes) over the area of forecast interest, separated from a coarser grid (4 km) near the boundaries by a variable resolution transition zone. This variable resolution approach allows the boundaries to be moved further away from the region of interest, reducing unwanted boundary effects on the forecasts.
UK ensemble forecasting, supported by a global ensemble, provides information on the uncertainty in forecasts up to five days ahead. The solution spread is obtained by perturbing both the initial conditions and also some aspects of the physical processes within the model. A medium-range global ensemble supports probabilistic weather forecasting out to seven days ahead.
Model |
Grid length in mid-latitudes |
Vertical levels |
Forecast length |
Run times (UTC) |
Initial conditions |
---|---|---|---|---|---|
Deterministic Global |
10 km (2560 x 1920 grid points) |
70 |
6 days |
00, 06, 12, 18 |
Hybrid Incremental 4D-Var with flow-dependent background errors from 44-member global ensemble. |
Ensemble Global |
20 km (1280 x 960 grid points) | 70 (lid ~80 km) |
Control member + 17 perturbed members to 7 days. 36 member ensemble generated by time-lagging over 12 hours (latest 2 cycles). |
00, 06, 12, 18 |
44-member ensemble of 4D-ensemble-Var. Model perturbations from Additive Inflation from historical analysis increments, Stochastic Kinetic Energy Backscatter (SKEB), Stochastic Perturbation of physics Tendencies (SPT), SST and soil-moisture perturbations. |
Deterministic UK (UKV) |
1.5 km inner domain (622 x 810 grid points); 4 km outer domain (950 x 1025 grid points) |
70 |
120 hours |
03,15 |
Hourly 4D-Var |
54 hours |
00, 06, 09, 12, 18, 21 |
||||
12 hours |
All other hours |
||||
Ensemble UK |
2.2 km inner domain; 4km outer domain (740 x 752 grid points) |
70 |
Control member + 2 perturbed members to 5 days |
05, 11, 17, 23 |
UKV analysis + perturbations interpolated from global ensemble. Lateral boundary conditions from corresponding global ensemble member. |
3 perturbed members to 5 days |
All other hours |
||||
18 member ensemble generated by time-lagging over 6 hours (latest 6 cycles) |
Every hour |
Crisis Area Models (CAMs)
The Met Office also has the capability to rapidly relocate regional models to any area of interest worldwide. These Crisis Area Models (CAMs) are run in support of allied military operations and disaster relief.
Differences compared to climate configurations
The weather forecasting configurations of the Unified Model differ from the Met Office seasonal and climate models in a number of respects:
- Data assimilation - In weather forecasting an accurate representation of the initial state of the atmosphere is vital and so assimilation of the latest observations is required.
- Resolution - The shorter forecast length allows the model to be run at higher resolution allowing more accurate representation of atmospheric dynamics and surface features.
- Coupling - While climate projections are typically run from coupled ocean-atmosphere models, the current NWP configurations use a fixed SST from the OSTIA analysis system. In the near future, however, we do plan to use a coupled ocean-atmosphere model for global NWP.
- Complexity - Processes which are not thought to have a significant impact on weather prediction timescales such as atmospheric chemistry and the carbon cycle are not currently modelled.